Communications · Article 3 of 10
AI email categorisation
When an email arrives, an AI reads it and assigns a category automatically. This classification helps the system apply the right tags, trigger the right automations, and give your team a quick signal about what kind of email they're dealing with.
The nine categories
| Category | What it means |
|---|---|
| Invoice | Submitting an invoice for services delivered |
| Quote | Providing a quote or cost estimate for future services |
| Receipt | A receipt or claim for reimbursement |
| Service agreement | Contains or relates to a service agreement document |
| Invoice follow-up | Sender is following up on invoice status |
| Junk | No business purpose — marketing, newsletters, automated notifications |
| General NDIS question | A question that can be answered without specific participant plan data |
| Specific NDIS question | A question that requires access to a specific participant's plan data |
| Other | Legitimate business communication that doesn't fit above |
An email can have more than one category
Navigator doesn't force a single category per email. If an email is both sending an invoice and asking a question about a participant's plan, it can be classified as both Invoice and Specific NDIS question.
This means the email can trigger the right responses for each aspect — routing to someone who processes invoices and flagging there's also a question to answer.
How the classification is used
The category feeds into two things:
- Tags — certain categories automatically trigger system tags. For example, an email classified as Invoice gets the
invoicetag. - Automation triggers — automation rules can use email tags (applied from categories) as conditions.
What to do if an email is miscategorised
AI classification is accurate most of the time but can occasionally get it wrong — especially for unusual formats or ambiguous content.
If an email has been miscategorised, an account administrator can reprocess the email to re-run the classification.
If you notice a pattern (e.g. a particular supplier consistently classified incorrectly), set up an automation rule to manually correct tagging for emails from that sender. See Email automations.